There are only two possible reasons this is lower on vauto. One is higher PS. The other is a lower trigger sensitivity. The ability to use higher EPR/PS and tweak the pressure wave form using timing controls are the only differences between these machines. Its the same blower motor operating at approximately the same pressures with approximately the same waveforms (minus what you tweak with timing controls).
These machines are much more basic then you give them credit for. They don't care nor take into account your chipmunk cheeks etc. Mask pressure and flow rate are simple calculations done at the machines. What you are questioning the accuracy of is flow rate as this is the data where you are seeing these fluctuations. Flow rate is calculated by the machine using some combination of sensors and I assume a vfd driven blower. Put on your mask, turn on the machine and place a hand near the inlet filter and breath in (you can feel air being sucked into machine) then breath out (you can feel air being pushed out). It is that flow through the machine that is being calculated/recorded and as far as I know could be as simple as recording the blower positioning. What is happening in your mouth, lungs etc does not matter, all that does is affect what pressures/flows are seen and recorded at the machine.
As a mechanical engineer I have been blown away by the accuracy of the flow rate data recorded by these machines. When I use my camera I can see a visible explanation for almost every odd breath I see in my OSCAR data. To review my camera data I literally look at the time of an odd shaped breath in OSCAR and then look at that time on the video. There is no point in watching the full 8+ hrs of video because most of it is me sleeping normally and almost every movement, odd breath or arousal shows up in my OSCAR flow rate data. I know because I did watch almost every minute of my first few videos when first comparing to OSCAR.
These flow rate charts are so accurate you can literally determine your heart rate from them by calculating the spacing between the extremely minor exhalations caused by expanding blood vessels in your lungs (aka determining the spacing of cardiogenic oscillations). This blew my mind and it wasn't until I saw my recent CT images that I was able to better understand why we can see these cardiogenic oscillations in our data. You can see something similar by clicking axial or coronal lung window in the viewer at this link and then scrolling up/down through images to see the blood vessel structures throughout the lungs and realizing that the vessels take up a fair amount of space. Cardiogenic oscillations in data are physical flow changes being measured at the machine, crazy accurate measurement ability.
https://radiopaedia.org/cases/normal-ct-chest?lang=us
Now stop and think about this for a bit. You have seen this on multiple machines, it is not a calculation error. We know these flow rate data charts are extremely accurate and it is something real being caused by something your body is physically doing. The big kicker is that this is not a little blip like a cardiogenic oscillation, your body is creating rapid, large amplitude air flow direction changes at the machine. I've already recommended video recording this multiple times but will do so again. Once you catch this on video it will be blatantly obvious what is happening (although maybe not blatantly obvious what is causing it), if it isn't obvious I will be very surprised.